It probably goes without saying in the cybernetic environment of the twenty-first century that effective epidemiology of all types, including pharmacoepidemiology, can only be seriously conducted with the addition to the armamentarium of the epidemiologist, of the skillful use of large, automated, multipurpose, population-based systems (the LAMPS)—known by shorthand as 'the databases'. Often these databases have been developed with a primary intent of creating economic efficiency, quality assurance, or management controls within organized systems of healthcare. Hence, in the USA, the organizations that construct these databases include insurance companies, hospitals, health maintenance organizations, and other companies in the healthcare business. In Canada, and increasingly in Europe, such databases are emerging from provincial/regional or national reimbursement programs. If the database is equipped with patient identifiers (e.g. a unique membership number), then hospitalizations, prescriptions and combinations of healthcare transactions can be linked to a single individual across components of the system and over time: a so-called 'record-linkage' system. More recently, the evolution of a powerful clinical management tool, the electronic medical record, further powers the availability of linked data for entire populations under care. Such databases render it feasible to assemble cohorts of drug-exposed individuals and computer-matched comparator populations from historical (extant) data and observe them (using cohort analyses) forward over the time in the database (often decades) for evidence of excesses of events under study. Similarly, case-control methods may assemble cases and comparators, and use the powerful databases as the source of the antecedent information, so elusive in hands-on methods.
Recent regulatory efforts on behalf of the needs to protect patient privacy are no strangers to this field, which has established a long and successful record of systems that protect patient privacy while assuring access to necessary population-level, individual-linked data. The recent, excellent policy positions on data privacy protections of the American College of Epidemiology (ACE) and the International Society for Pharmacoepidemiology (ISPE) stand as evidence of this competence. The reader is referred to the websites of these organizations.
It is to be emphasized that such database work is often complicated, and requires a team of professionals comprising physician and non-physician pharmacoepidemiologists, statisticians, and specialists in information technology. Perhaps one of the greatest contributions that a clinician can make to such a team is to provide relevance to the hypotheses that are tested and as a reality check on the results that the computers generate, and which those less close to the field tend to regard automatically as 'fact'.
Despite the deserved enthusiasm for the contribution of the LAMPS to epidemiology, more traditional hands-on, structured observational studies, with enrollment of cohorts of persons exposed to an agent under study and proper comparator populations, and selection of cases (e.g. from medical records) and appropriate controls, still have specific applications in pharmaceutical medicine, thus characterizing part of the activity of pharma-coepidemiologists.
Was this article helpful?